Expedia deploys chatbots and machine learning as it monitors every customer click

Expedia uses every single click a customer makes on its website as it looks to build up a profile of their travel needs and is using the data – alongside machine learning – to “optimise the customer experience”. After the online travel agent announced its new usability lab on Tuesday morning, vice president in product … Continue reading Expedia deploys chatbots and machine learning as it monitors every customer click

Expedia uses every single click a customer makes on its website as it looks to build up a profile of their travel needs and is using the data – alongside machine learning – to “optimise the customer experience”.

After the online travel agent announced its new usability lab on Tuesday morning, vice president in product management Scott Crawford explained how Expedia’s new Scratchpad service is designed to help customers get the best deal from 19,000,000,000,000 (19 quadrillion) possible searches.

Scratchpad stores customer’s searches, and saves them on their Expedia account in what Crawford called a “more elegant” way than scribbling down reams of notes on paper about hotels, flights, transfers and extras (according to Crawford, the average flight shopper looks at 48 different options before they book).

It tracks prices as they change in real-time, and even tells you if you might be able to save money by waiting to book in the future.

“People are looking for different outcomes, but everyone wants a great holiday,” Crawford said. “The question is, how do we best match travellers for the optimal outcome for them?”

Expedia has vastly increased the amount of tests it runs – from 50 per year in 2010 to 1,500 in 2016. In that time it has invested $2 billion in technology.

The data it has gleaned amounts to 6.7 billion 200-page books, 189 billion rows of data and Expedia employs 100s of scientists and analysts to interpret it.

“Scale gives us an advantage in innovation,” said Crawford. “The quicker we can test, the quicker we can learn.

“Every purchase, every interaction, every click is logged, we monitor price, location, reviews and personalise the experience.

“Every time someone goes back, the site gets better for that person. The more data we can get, the quicker we can optimise the algorithm.”

Customers are also contacted during their stay to offer ‘real-time reviews’ while they are in a hotel or just come off a flight and Expedia has also introduced its ‘search anything’ option.

Still in its infancy, the idea is that the more patterns Expedia’s computers see, the more signals it gets and can tailor-make your holiday for you.

The conversational style has echoes of a traditional travel agent.

Expedia is also looking into the use of chatbots such as Facebook Messenger, using natural language recognition to develop more of that conversational feel.

“We see a lot of opportunity here,” said Crawford. “We have invested a lot and are excited to see where it takes us.”

He added that Expedia works with more than 43,000 travel agents worldwide.